Skip to content

Podcast

How to Build the AGI Future- Bob McGrew

Y Combinator Startup Podcast

Source ↗ ← All highlights
  • OpenAI Journey
    • Bob McGrew joined OpenAI to learn deep learning after realizing robotics wasn’t ready for startups.
    • He taught a robot checkers and learned about scaling AI through projects like Dota 2 and Rubik’s Cube. Transcript: Garry Tan You know, what was that like early at OpenAI? Bob McGrew The really interesting thing about OpenAI is that I did not originally intend to go to a research lab. When I left Palantir, I wanted to start a company. I had a thesis that robotics would be the first real business that was built out of deep learning. This was back in 2015. And I talked my way into a friend’s nonprofit. I never had a badge, but I would go in, he’d open the door for me. And I learned deep learning by teaching a robot how to play checkers from vision. And in the process of doing this, I learned a lot about robotics and I learned that robotics was definitely not the right startup to start in 2015 or 2016. I ended up going to open AI basically because it was a place full of very smart people and it had, you know, big ambitions. It was a place where I could really learn. I had all this management experience from Palantir, but it was just a place for me to really become an expert in deep learning. And, you know, from there, figure out what it could actually be used and applied for. (Time 0:00:58)
  • Early OpenAI Strategy
    • OpenAI’s early strategy involved research and papers, but they recognized its limitations.
    • Projects like Dota 2 and the Rubik’s Cube hand reinforced the importance of scale in AI. Transcript: Garry Tan Were some of the earliest things that you remember working on? And how did that play into what everyone knows OpenAI to be now? Bob McGrew Yeah. When OpenAI started, the goal was always to build AGI. But the theory early on was that we would build AGI by doing a lot of research and writing a lot of papers. And we knew that this was a bad theory. I think for a lot of the early people who were startup people, Sam, Greg, myself, it felt sort of painful and a little academic. But at the same time, it was sort of what we could do at the time. And so some of the early projects, I worked on a robotics project where we took a robot hand, a humanoid robot hand, and we taught it to solve Rubik’s Cube. The idea in doing that was that if we could make the environments complicated enough, the artificial intelligence would be able to generalize out of the narrow domain it was taught And learn something more complicated, which was one of the ideas that later we see coming back with LLMs. The other really early big project was solving Dota 2. So there’s a long history of solving games as a path towards building better AI from Othello to Go. And after beating Go, the next hardest set of games are actually video games. They’re not very classy, but they’re a lot of fun. And I can assure you that mathematically they were harder. And so DeepMind went after StarCraft, OpenAI went after Dota 2. And there was real insight that was generated there, which was that it really strengthened our belief that scale was the path to improving artificial intelligence. That with Dota 2, the secret idea was that we could take huge amounts of experience and feed it into a neural network and that the neural network would actually learn and generalize from That. And later we actually went back and applied this to the robot hand and that became the key idea for the robot hand. And at the same time as these two big projects were going on, Alec Radford was experimenting with language. And the core idea behind GPT-1 is that if you have a transformer and you apply this super simple objective of guessing the next token, guessing the next word, that that would be enough Signal that you could actually have something that would be able to generate coherent text. And in retrospect, it sounds sort of obvious, right? Like, you know, clearly this was going to work, but no one thought this would work at the time. Alec, you know, really had to persevere for years in order to make this work, and that became GPT-1. And then after GPT-1 seemed successful, we brought in the ideas from Dota and from the robot hand of training at larger and larger amounts of scale and training on a really diverse set Of data and looking for generalization. (Time 0:02:04)
  • OpenAI Culture
    • OpenAI’s culture differed from DeepMind and Google Brain, focusing on scaling ideas.
    • They valued internal reputation over paper credit, fostering collaboration and scale. Transcript: Garry Tan And GPT-4. So one of the things that your OpenAI really pioneered and sort of figured out was this concept of scale. How is it that it was OpenAI that, you know, made the right decisions and sort of found, you know, large language models first? Bob McGrew Early on, there were sort of, you know, a couple big projects, as I said, and then some room for exploratory research. And at the very earliest days, the exploratory research was really about, you know, it’s about what the researcher wanted to do. But also, it was about sort of the company’s opinion. And in this, it was primarily formed by ILIO with influence from a lot of people. But I think Ilya was really the guiding light here early on. Sometimes I think about the open AI culture and I like to oppose it to sort of Google brain and to deep mind. And so early on, the deep mind culture was, you know, a caricature is Demis had a big plan and he wanted to hire a bunch of researchers so he could tell them to move forward with his plan. And Google brain said, let’s rebuild academia. Let’s like bring in all these super super talented researchers. Let’s not tell them anything. Let’s just let them figure out what they want to do, give them lots of resources, and hope that amazing products pop out. And of course, they did, but they didn’t necessarily happen at Google. And we took a different approach, which was really more like a startup, where there was no sort of big centralized plan. But at the same time, people didn’t have, it wasn, it wasn’t just sort of let’s let a thousand flowers bloom. Instead, we had opinions about what needed to be done and things like how do you show scale as a way of, you know, making your idea get better. And that opinion was set by the research leadership. You know, again, early on, people like Ilya, people like Dario. That was how we made sure that we didn’t just sort of throw resources at everybody, but neither did we have just one set of ideas that were there. (Time 0:04:50)
  • Scaling Laws in AI
    • Scaling laws are prevalent in AI, impacting how companies and researchers should approach development.
    • Building the first working model is crucial before applying scaling laws, as seen with DALL-E. Transcript: Garry Tan You have the scaling laws, and certainly how AI research is being done now, there’s sort of this shift basically scale is all you need for increasingly more and more AI domains. It’s sort of potentially coming true in image diffusion models or in earlier, to bring it back to what you’re starting out with. There’s some sense that similar principles to scaling laws actually do apply in the right domains in robotics. Is that sort of one of the things that you’re seeing or how would you respond to that? Bob McGrew I think if you look at AI progress, you see scaling laws all over the place. And so the interesting question is, well, if scaling laws exist and they’re commonplace, what does that mean? What does that mean for you if you’re a company, if you’re a researcher, if you’re trying to make things better? Garry Tan Why didn’t we take advantage of scaling laws earlier in these other domains? Bob McGrew Well, you know, I think we were really trying to. You know, usually in order to, the first step is actually getting to a scaling law. To take an example that’s not LLMs. If you think about DALI, which was, you know, how do you take text and make an image out of it? That, I think Aditya Ramesh, who built that model, spent 18 months, maybe two years, just getting to the first version that clearly worked. So I remember he’d be working on this and Ilya would come and show me. He’d be like, you know, Aditya’s been working on this for a year. He’s trying to make a pink panda that’s skating on ice because it’s something that’s clearly not in the training set. And here’s an image and you can see it’s like pink up there and white down there. It’s really beginning to work. And I would look at that and I’d be like, really? I mean, maybe, maybe, I don’t know. But just getting to that point where it sort of plausibly begins to work is a huge, difficult problem. And it’s completely separate from using scaling laws. Now, once you get it to work, that’s when scaling laws come into play. And with scaling laws, you have two hard things that you can do. One of them is just the pure scale itself. Scaling is not easy. It is in fact probably the practical problem in any sort of model building. And it’s a systems problem. It’s a data problem. It’s an algorithmic problem, even if you’re just trying to scale the same architecture. The second thing you can do is you can try to change the slope of the scaling law or just bump it up a little bit. And that is searching for better architectures, searching for better optimization algorithms, all of the algorithmic improvements that you can do. And if you put all of those together, that is what explains the very fast progress that we’re seeing in AI today. (Time 0:08:42)
  • Beyond Scaling Laws
    • Scaling laws may hit data limits, but reasoning and test-time compute offer new avenues for advancement.
    • Reasoning unlocks agent capabilities and reliable action-taking. Transcript: Garry Tan I guess that is one of the bigger debates that’s ongoing, certainly out there in the community. Are the scaling laws going to continue to hold or are we hitting some sort of bottlenecks? I don’t know how much you can talk about it, but what’s your view at this point on maybe LLM scaling, but certainly other domains too? Bob McGrew It is definitely the case that there is a data wall and that if you take the same techniques that we were using to scale LLMs, at some point you’re going to run into that. The thing that’s been really exciting, of course, is going from the LLM scaling of pre-training where you’re just bringing bigger and bigger corpuses and trying to predict the next Token and shifting gears and using techniques like reasoning which you know open ai shipped in its 01 and 03 models and gemini has now also shipped in gemini flash thinking you know if You think about moore’s law right you know with moore’s law you see you know the moore’s law is sort of one big exponential curve, but it’s actually the sum of a bunch of little S curves. And you start off with Dennard scaling. And at some point that breaks. But if you look, if you think about how NVIDIA has gone, Moore’s law has continued. It’s just come through a different mechanism. Garry Tan So you like solve some bottleneck, but then you S curve that particular solution, But there are other places where there are other bottlenecks. Bob McGrew And then you have a new bottleneck and you have to go attack that. And so, you know, I think in the big picture, we’re reaching that bottleneck for pre-training and data. Are we exactly there? It’s a little hard to say, but now we have this new mechanism with reasoning and test time compute. I think if you go back and you think about what it took for AI, for building AGI, for, you know, I would say the last five years, people have thought that, people at the big frontier labs Have felt that, you know, step one was pre-training and that the remaining gap to have something that could scale all the way to AGI was reasoning. Some ability to take the same pre-trained model and have the ability to give it more time to think or more compute of various kinds and get a better answer at the other end. And now that that has been cracked, at this point, I think we actually have a very clear path to just focus on scaling. We were talking about that, the zero to one part that’s not about scaling. I think there’s a really strong case to be made that in LLMs, that’s not relevant anymore. (Time 0:11:15)
  • Advice for AI Startups
    • AI startups should start with the best model possible, focusing on frontier AI capabilities.
    • Distillation and cost optimization come after identifying value through user iteration. Transcript: Garry Tan Would you say to people watching who are trying to make AI startups right now? Often they’re vertical startups, but some of them are consumer too, actually. Bob McGrew Yeah, I would say if you’re a founder, the right approach is to start with the very best model you can, because your startup is only going to be successful if it exploits something about AI that realistically is going to be on, you know, the frontier. So start with the very best model that you can and get it to work. And once you’ve gotten it to work, then you can use distillation. You can take a dumber model and you can try prompting it. You can try to have the frontier model, train the smaller model. But, you know thing in a startup is actually your time. Unless you have to, you don’t want to be like Palantir taking three years to get to market. You want to be able to build that product as quickly as possible. (Time 0:17:34)
  • Slow AI Adoption
    • AI adoption is surprisingly slow, despite capabilities that were predicted to automate many jobs.
    • Personalized AI assistants, deeply integrated with user context, are a promising market gap. Transcript: Bob McGrew Do you want to do it? Yes or no. Right. And there’s something really interesting about this idea because I think it’s very compelling that the AI is your life coach. But then it goes back to like, so what are you even doing with your life in the first place, Right. If the AI is better than you. And I think there’s actually a really deep mystery here. When we were first thinking about GPD one back in 2018, you know, if you ask people what AGI was, they would say, well, you know, it’s, it’s, you know, a model that you can actually interact With. It passes the Turing test. It can look at things. It can write code. It can even draw an image for you. We’re there. Yeah. And like, we’ve had this for years, right? And if you said, okay, well, what happens when you get all those capabilities? Say, well, everybody’s out of a job. You know, all laptop jobs are immediately automated and game over for humanity. And none of that is happening, right? I mean, yes, AI has had some effects, you know, particularly on people who write code. But, you know, I don’t think you can see it in the productivity statistics, unless it’s about how big the data centers are that we’re building. And I think this is a really deep mystery. Why is it that AI adoption is so slow relative to what we thought should be happening in 2018. (Time 0:20:00)
  • Palantir’s Forward-Deployed Engineers
    • Palantir’s success with forward-deployed engineers shows the value of specialized software.
    • AI needs software engineers to bridge the gap between intelligence and user needs. Transcript: Garry Tan What you just said really reminds me of our days at Palantir, actually, where one of the core missions that Palantir started with, really, is this idea that the technology is already Here. It’s just not evenly distributed. And I feel like that was one of the things you guys actually really discovered. And part of the reason why Palantir actually exists, you went into places in government, three letter agencies, some of the most impactful decisions that a society might have to make. And you look around and there was no software in there. And that was what Palantir and certainly Palantir government was very early on. Bob McGrew The fun piece there was just, you know, thinking through what it is that these people do and then how you could just completely reimagine it with technology. Where, you know, if you were checking to see if a particular, you know, person who was flying into the U.S. Had a record or if there was any suspicion, you know, you look through 20 different databases. One approach would be say, well, let’s make it faster to look through 20 different databases. Another approach is say, well, maybe you can just look for it once and it checks all the databases for you. And I think we need some twist like that for AI that lets people figure out how to use the AI to solve the problem they actually have, not just sort of take their existing workflow and have AI do that workflow. Yeah. Garry Tan It’s like not just having the data. It’s not just having the intelligence. I mean, what AI desperately needs right now is, like you said, the UI, the software. It’s just building software. And if you can put that in a package that a particular person really, really needs, I feel like that’s one of the big things that we learned at Palantir. It’s like, there’s a whole job that is exactly that, forward deployed engineer. It’s a very evocative term, right? Like forward deployed, you’re not way back at the HQ. You’re all the way in the customer’s office. You’re sitting right next to them at their computer watching how they do something. And then you’re making the perfect software that they would never get access to. The alternative is Excel spreadsheet, writing SQL statements yourself, or Cost Plus, government integrator, or Accenture, and they’re never going to get something usable. Whereas a really good engineer who’s a good designer who can understand exactly what that person needs and is trying to do, they can build the perfect thing for that very person. And so maybe that’s the answer to your question. Like, why didn’t it happen yet? It’s like, we just need more software engineers who are like that forward deployed engineer to link up the intelligence and we’re there. (Time 0:21:10)
  • Future of Work
    • Bob McGrew uses language models to create coding lessons for his son, promoting critical thinking.
    • Future jobs will involve lone geniuses leveraging AI and managers leading AI-driven teams. Transcript: Garry Tan Both of us are parents and, you know, we just spent a lot of time talking about some pretty wild concepts that are about to affect all of society. Has that affected how you think about, you know, what we should be doing with our kids? Bob McGrew I really struggle with this. And there’s a very crisp version of this for me, which is that my eight-year son is really excited about coding. He actually is really excited. He wants to start a company. He has a great name and it’s going to do asteroid mining and all sorts of cool stuff. And so every day he says, you know, dad, can you teach me a little bit about how to code? This is actually what I do most with language models is I have the language model. I figure out what he’s interested in. I have the language model, make a lesson for him that teaches some idea that I want to teach him. Like it teaches him about networking or teaches him about loops and it fits his idea. And my wife asked, why are you doing this if language models are going to be able to code? And I think the answer is that right now, this is how you learn how to do critical thinking. And I think back to Paul Graham’s idea of the resistance of the medium. Even once the computer can do the programming for you, I think there’s still something to to having like had your hands in it yourself and knowing what’s possible and what’s not possible And that you can have that intuition i think that the the role that we’re going to be playing you know one i i think there’s going to be two roles one will be something like alone genius you Know the alec radford of the world working alone at his computer, coming up with some crazy idea. But now with that computer being able to leverage him up so much. And the other role is manager that, you know, you will be the CEO of your own firm and that firm will mostly be AI. I think it will be other humans in there. I don’t think the whole company gets replaced, although this is another really interesting question for us to answer. But, you know, I think those will be the two jobs of the future. Genius and manager. (Time 0:24:40)
  • Future of Robotics
    • Robotics companies are currently where LLMs were five years ago, poised for significant growth.
    • Scaling in robotics is harder due to physical constraints, but foundation models show promise. Transcript: Garry Tan I guess going back to robotics, you know, one of my hopes is actually that maybe the level four innovators will suddenly break through on a bunch of very specific problems that currently Hold back robotics. Have you spent time back in that space recently? And what are the odds of that coming together in the next, I don’t know, a couple of years even? Like, do you feel like there will be continued breakthroughs on maybe the figure robot and different things like that? What’s your sense for robotics in the next year or two? Bob McGrew Robotics companies now are where, you know, LLM companies were five years ago. So I think in five years, you know, or even sometime in the next five years, we will see the chat GPT moment for robotics. I think it’s a little harder to scale because you’ve got to build physical robots. But if you look at companies like skilled AI or physical intelligence, who are building foundation models for robots, you know, the progress that we’ve seen there is just really dramatic. There’s some point we’re going to get out of that zero to one phase where you’re just trying to make it work at all. And we’re going to be in something where it kind of works. And then we’re just scaling to increase the reliability and increase the scope of the market. (Time 0:27:56)